The proposed interdisciplinary trial will test a tailored intervention to increase self-efficacy for conducting research in minority and non-minority undergraduate and graduate students. The study is designed to evaluate group differences in perceived self-efficacy for research, same and mixed gender dyads, minority and non-minority dyads, as well as importance attached to characteristics and behaviors of research role models, and students' intentions regarding career choices. The theoretical framework of the intervention is derived from social cognitive theory (Bandura, 1986; 1997). The theory-based intervention will be delivered to mentors and their prot[unreadable]g[unreadable]s to increase prot[unreadable]g[unreadable]s' perceived ability to perform various tasks and activities in the research domain. Targeted subjects will include mentor-prot[unreadable]g[unreadable] dyads consisting of 183 university researcher mentors currently receiving extramural grant funding, and 183 prot[unreadable]g[unreadable]s (student research assistants) working on the mentors' funded projects. Based on power calculations, and allowing for attrition, with this sample size we will have 84% power to detect a difference in scores. The study will employ a repeated measures experimental design. One hundred eighty-three dyads will be randomized to three groups: (1) a face-to-face intervention group (n=61 dyads); (2) an Internet intervention group (n=61 dyads); and (3) a no-contact control group (n=61 dyads); at least 150 (50 per group) are expected to complete data collection. Data collected at the outset will include biographic data forms from mentors and prot[unreadable]g[unreadable]s and attributes of mentors. Data, collected on three occasions (pre-intervention, 3 months, and 6-months after intervention), consists of perceived self-efficacy for research of the prot[unreadable]g[unreadable]s, prot[unreadable]g[unreadable] career goal statements, and weekly prot[unreadable]g[unreadable] logs of time spent and activities shared with mentors. [unreadable] [unreadable] [unreadable]